@mcpher/gas-fakes 2.3.17 → 2.3.18

package/gf_agent/SKILL.md

@@ -164,8 +164,9 @@ Agent:
 
 
 ### Google Sheets (SpreadsheetApp)
-- **Chart Creation & Ranges**: When using `EmbeddedChartBuilder.addRange()`, the Sheets API requires `ChartSourceRange` domains and series to have a length of 1 for either rows or columns.
-  - **Crucial**: Do **not** pass a multi-column range to `addRange()`. Add domains and series as separate single-column ranges.
+- **Efficiency & Batching (CRITICAL)**: Always prefer `range.setValues(array)` over making multiple individual `range.setValue(val)` calls. If you need to write multiple rows or columns of data to a sheet, batch them up into a 2D array and write them all at once. This significantly improves performance and avoids unnecessary API rate limits.
+- **Chart Creation & Ranges**: While `gas-fakes` now emulates Live Apps Script behavior by automatically splitting multi-column ranges in `addRange()` (first column as domain, rest as series), for **maximum reliability** and clear control over your chart structure, it is still recommended to add domains and series as separate single-column ranges.
+  - *Example*: `chart.addRange(sheet.getRange("A2:A10")).addRange(sheet.getRange("B2:B10"))` is preferred over `chart.addRange(sheet.getRange("A2:B10"))`.
 - **Values vs. Display Values**: 
   - `getValues()` returns unformatted data.
   - `getDisplayValues()` returns formatted strings.
@@ -409,7 +410,9 @@ Unlike the standard Apps Script Services (`SpreadsheetApp`, `DriveApp`), the sig
 If you are orchestrating a complex task that requires an Advanced Service (such as resizing an image via `Docs.Documents.batchUpdate` or applying granular formatting via `Sheets.Spreadsheets.batchUpdate`) and you do not know the exact JSON payload structure, you MUST research it using the Google API Discovery documents.
 
 **How to Research Advanced Services:**
-Do not guess the payload structure. Instead, use the `run_shell_command` tool to `curl` and `grep` the official Discovery Document for the specific API version.
+You SHOULD FIRST use the `run_shell_command` tool to `curl` and `grep` the official Google API Discovery Document for the specific API version. This is the most reliable way to guarantee accurate, 100% current REST JSON schemas.
+
+**Web Search Fallback**: Only if `curl`ing the Discovery API fails, or if it does not provide enough clarity to construct the request, you MAY use the `google_web_search` tool as a fallback default. Be cautious: web search often returns outdated or non-REST API examples (like Java or Python SDKs) which cause script failures.
 
 **Discovery Document URLs:**
 - **Docs API v1**: `https://docs.googleapis.com/$discovery/rest?version=v1`
@@ -423,6 +426,10 @@ If you need to know how to structure an `insertInlineImage` request for the Docs
 ```bash
 curl -s "https://docs.googleapis.com/$discovery/rest?version=v1" | grep -A 30 '"InsertInlineImageRequest":'
 ```
+For `deleteObject` (e.g., deleting an image):
+```bash
+curl -s "https://docs.googleapis.com/$discovery/rest?version=v1" | grep -A 20 '"DeleteObjectRequest":'
+```
 
 By fetching the exact schema from the discovery document, you ensure your `batchUpdate` arrays and payload objects are 100% accurate before generating the execution script.
 
@@ -441,6 +448,23 @@ When generating code that uses the `Utilities` service, you must adhere to the f
    - If you are writing tests or robust `try/catch` blocks intended to run cross-platform, DO NOT assert against the exact string value of the error message (like `e.message.includes("failed")`). Simply check that an exception was thrown.
 
 
+### Context Efficiency & Logging (CRITICAL)
+
+To maintain a clean and professional user experience, `gf_agent` MUST prioritize context efficiency and minimize redundant tool logging.
+
+1. **Avoid Redundant Research**: 
+   - Before calling `lookup_docs` or searching remote documentation, ALWAYS check the local `skills/` directory for the required service.
+   - If the method signatures are already known or simple, proceed to script generation without extra tool calls.
+   - DO NOT call `lookup_docs` for every service in every turn. Only call it when a specific method signature is unknown or when a script fails with a "not a function" error.
+
+2. **Minimize Output Verbosity**:
+   - When reporting the results of research to the user, DO NOT print long lists of method names found in documentation. Summarize only the relevant findings.
+   - The user does not need to see the "raw" output of discovery tools.
+
+3. **Quiet Execution**:
+   - Aim for a "one-shot" success pattern. Use the gathered knowledge to write a robust script that works on the first try, avoiding the "Retry/Correction" cycle which generates excessive terminal logs.
+
+
 # gf_agent Knowledge Base
 
 This directory contains modular markdown files representing the "Lessons Learned & Best Practices" for the `gf_agent` skill.

package/gf_agent/knowledge/05-sheets-forms.md

@@ -1,6 +1,7 @@
 ### Google Sheets (SpreadsheetApp)
-- **Chart Creation & Ranges**: When using `EmbeddedChartBuilder.addRange()`, the Sheets API requires `ChartSourceRange` domains and series to have a length of 1 for either rows or columns.
-  - **Crucial**: Do **not** pass a multi-column range to `addRange()`. Add domains and series as separate single-column ranges.
+- **Efficiency & Batching (CRITICAL)**: Always prefer `range.setValues(array)` over making multiple individual `range.setValue(val)` calls. If you need to write multiple rows or columns of data to a sheet, batch them up into a 2D array and write them all at once. This significantly improves performance and avoids unnecessary API rate limits.
+- **Chart Creation & Ranges**: While `gas-fakes` now emulates Live Apps Script behavior by automatically splitting multi-column ranges in `addRange()` (first column as domain, rest as series), for **maximum reliability** and clear control over your chart structure, it is still recommended to add domains and series as separate single-column ranges.
+  - *Example*: `chart.addRange(sheet.getRange("A2:A10")).addRange(sheet.getRange("B2:B10"))` is preferred over `chart.addRange(sheet.getRange("A2:B10"))`.
 - **Values vs. Display Values**: 
   - `getValues()` returns unformatted data.
   - `getDisplayValues()` returns formatted strings.

package/gf_agent/knowledge/13-advanced-services-discovery.md

@@ -5,7 +5,9 @@ Unlike the standard Apps Script Services (`SpreadsheetApp`, `DriveApp`), the sig
 If you are orchestrating a complex task that requires an Advanced Service (such as resizing an image via `Docs.Documents.batchUpdate` or applying granular formatting via `Sheets.Spreadsheets.batchUpdate`) and you do not know the exact JSON payload structure, you MUST research it using the Google API Discovery documents.
 
 **How to Research Advanced Services:**
-Do not guess the payload structure. Instead, use the `run_shell_command` tool to `curl` and `grep` the official Discovery Document for the specific API version.
+You SHOULD FIRST use the `run_shell_command` tool to `curl` and `grep` the official Google API Discovery Document for the specific API version. This is the most reliable way to guarantee accurate, 100% current REST JSON schemas.
+
+**Web Search Fallback**: Only if `curl`ing the Discovery API fails, or if it does not provide enough clarity to construct the request, you MAY use the `google_web_search` tool as a fallback default. Be cautious: web search often returns outdated or non-REST API examples (like Java or Python SDKs) which cause script failures.
 
 **Discovery Document URLs:**
 - **Docs API v1**: `https://docs.googleapis.com/$discovery/rest?version=v1`
@@ -19,5 +21,9 @@ If you need to know how to structure an `insertInlineImage` request for the Docs
 ```bash
 curl -s "https://docs.googleapis.com/$discovery/rest?version=v1" | grep -A 30 '"InsertInlineImageRequest":'
 ```
+For `deleteObject` (e.g., deleting an image):
+```bash
+curl -s "https://docs.googleapis.com/$discovery/rest?version=v1" | grep -A 20 '"DeleteObjectRequest":'
+```
 
 By fetching the exact schema from the discovery document, you ensure your `batchUpdate` arrays and payload objects are 100% accurate before generating the execution script.

package/gf_agent/knowledge/15-logging-efficiency.md

@@ -0,0 +1,15 @@
+### Context Efficiency & Logging (CRITICAL)
+
+To maintain a clean and professional user experience, `gf_agent` MUST prioritize context efficiency and minimize redundant tool logging.
+
+1. **Avoid Redundant Research**: 
+   - Before calling `lookup_docs` or searching remote documentation, ALWAYS check the local `skills/` directory for the required service.
+   - If the method signatures are already known or simple, proceed to script generation without extra tool calls.
+   - DO NOT call `lookup_docs` for every service in every turn. Only call it when a specific method signature is unknown or when a script fails with a "not a function" error.
+
+2. **Minimize Output Verbosity**:
+   - When reporting the results of research to the user, DO NOT print long lists of method names found in documentation. Summarize only the relevant findings.
+   - The user does not need to see the "raw" output of discovery tools.
+
+3. **Quiet Execution**:
+   - Aim for a "one-shot" success pattern. Use the gathered knowledge to write a robust script that works on the first try, avoiding the "Retry/Correction" cycle which generates excessive terminal logs.

package/package.json

@@ -40,7 +40,7 @@
   },
   "name": "@mcpher/gas-fakes",
   "author": "bruce mcpherson",
-  "version": "2.3.17",
+  "version": "2.3.18",
   "license": "MIT",
   "main": "main.js",
   "description": "An implementation of the Google Workspace Apps Script runtime: Run native App Script Code on Node and Cloud Run",

package/src/services/spreadsheetapp/fakeembeddedchartbuilder.js

@@ -59,15 +59,40 @@ export class FakeEmbeddedChartBuilder {
     if (nargs !== 1) matchThrow();
 
     const gridRange = makeSheetsGridRange(range);
+    const startCol = gridRange.startColumnIndex || 0;
+    const endCol = gridRange.endColumnIndex || startCol + 1;
+    const numCols = endCol - startCol;
 
-    if (this.__apiChart.spec.basicChart.domains.length === 0) {
+    // If it's a multi-column range, live GAS splits the first column as domain 
+    // and the rest as series (unless it's already got a domain)
+    if (numCols > 1 && this.__apiChart.spec.basicChart.domains.length === 0) {
+      // First column is domain
+      const domainRange = JSON.parse(JSON.stringify(gridRange));
+      domainRange.endColumnIndex = startCol + 1;
       this.__apiChart.spec.basicChart.domains.push({
-        domain: { sourceRange: { sources: [gridRange] } },
+        domain: { sourceRange: { sources: [domainRange] } },
       });
+
+      // Rest are series
+      for (let i = 1; i < numCols; i++) {
+        const seriesRange = JSON.parse(JSON.stringify(gridRange));
+        seriesRange.startColumnIndex = startCol + i;
+        seriesRange.endColumnIndex = startCol + i + 1;
+        this.__apiChart.spec.basicChart.series.push({
+          series: { sourceRange: { sources: [seriesRange] } },
+        });
+      }
     } else {
-      this.__apiChart.spec.basicChart.series.push({
-        series: { sourceRange: { sources: [gridRange] } },
-      });
+      // Standard behavior
+      if (this.__apiChart.spec.basicChart.domains.length === 0) {
+        this.__apiChart.spec.basicChart.domains.push({
+          domain: { sourceRange: { sources: [gridRange] } },
+        });
+      } else {
+        this.__apiChart.spec.basicChart.series.push({
+          series: { sourceRange: { sources: [gridRange] } },
+        });
+      }
     }
     return this;
   }